IOTIVITY AND EMBEDDED LINUX SUPPORT Kishen Maloor Intel Open Source Technology Center
Outline Brief introduction to IoTivity Software development challenges in embedded Yocto Project and how it addresses these challenges Key takeaway: IoTivity over Yocto makes an ideal platform for developing embedded IoT applications This is not a tutorial on Yocto
What is IoTivity? Internet Of Things Interconnecting physical objects with the digital world Widespread deployment of Low Power Embedded computers IoTivity High-level APIs for IoT Application Developers Exposing things as resources Discovering and manipulating resources over multiple network transports Utilize emerging IoT technologies
Simple Use Cases Turn Lights ON Light bulbs with BLE radios Smartphone Notify Current Setting 75F 75F Smart TV Digital Thermostat Regulate Temperature Tablet
IoTivity Software Stack CoAP coap://<deviceaddress>/temperature IoTivity Stack on an edge device OS Middleware Application C/C++ APIs IoTivity Resource Model & Services Connectivity Abstraction Network Interfaces User space Kernel space Thermostat Protocols and drivers
Emerging Open IoT Protocols 6LoWPAN: IPv6 over Low Power Wireless Personal Area Networks Bluetooth Smart IPSP RPL: Routing over Low Power and Lossy Networks New RFCs being published followed by prototype Linux implementations Growing influence of Linux in IoT
Challenges Heterogeneous nature of targets, CPUs, kernels IoTivity needs to be ported to each and maintained separately. Not easily scalable. IoT rapidly evolving with new protocols Need modular approach to quickly plug-in new IoT protocol implementations
Challenges Embedded development now becoming mainstream with IoT Need cohesive software development infrastructure that is uniform across multiple IoT targets These challenges are addressed by the Yocto Project
Yocto Project http://www.yoctoproject.org/ Hosted at the Linux Foundation Create customized OS images for embedded targets Ready-to-use BSPs for multiple platforms Layer-based flexible build architecture Focus on configurability and reuse Support for major CPU architectures
Yocto Build Workflow Target Machine Definition BSP OS Image Metadata For Builds And Patches Recipes For Software Components BitBake Build Task Executor Binary And Development Packages Package Feeds SDK Configuration
BitBake Recipes Represents a meta package Define contents of binary and development packages Dependency relationships between recipes Versioning Interfaces for fetch, patch, configure, compile, install steps Architecture specific switches
Software Layers Related collections of recipes to build applications and middleware Customize build and configuration of BSP and other software layers Package up IoTivity and dependencies in a target agnostic way
meta-oic Software Layer git://git.yoctoproject.org/meta-oic Samples IoTivity Resource clients and servers Third-party protocol plug-ins APIs Service Model and Plug-in Manager Resource Model Base Framework Dependencies Kernel Configuration Protocol implementations Middleware Updates
Kernel Builds In Yocto linux-yocto Upstream kernel based trees maintained by the Yocto Project Platform specific branches Recipes for respective kernels linux-yocto-custom Build any git-based kernel Use Yocto provided kernel tooling for build / configuration
Adding Kernel Features Create a.bbappend to build upon existing kernel recipes Resides in your layer and distributable Configuration Fragments Create a.cfg and place in your kernel.bbappend #Enable features for IoTivity CONFIG_BT_6LOWPAN=y CONFIG_IEEE802154=m CONFIG_IEEE802154_6LOWPAN=m CONFIG_6LOWPAN_IPHC=m CONFIG_MAC802154=m
Other Supporting Features Distribute new features as patches Middleware Adding a GATT interface for IoTivity to BlueZ Create a.bbappend for the BlueZ recipe Protocol integration RPL (Routing protocol for Low Power and Lossy Networks) XBee module for 802.15.4 support Security related features Opportunity to pack in early implementations of IETF specs via kernel patches
Application Development Application Development Toolkit Standalone cross-compiling toolchain with debugging and profiling tools Constructing an SDK Picks all development packages for target ADT will include IoTivity SDK Generates target ADT for specified build machine architecture IoTivity developers can focus on application development without getting bogged down by details of target
Yocto Eclipse Plug-in Eclipse integration with Yocto ADT Access to cross-compiler, debugging and profiling tools Remote application debugging, step through code Real hardware via network using its IP address QEMU Install Plug-in and point it to your target ADT Configure remote connection in C++ Remote Application under Debug Configurations
Remote Debugging
Releasing Your Application Write a BitBake recipe to build your application in the Yocto environment Distribute application packages for specific target platforms
Putting It To Test Built IoTivity and toolchains for Intel Edison, Galileo and MinnowBoard MAX BSPs available online C++ MinnowBoard/Edison applications built with ADT Android UI IoTivity MinnowBoard Application IoTivity Yocto Resource Aggregation Edison Ambient Light Temperature Application Sensing / Control IoTivity Yocto
To Conclude Yocto provides for greater scale Configure in one place, deploy on any Yocto-based platform Improved embedded IoT app developer experience Linux supports state-of-the-art IoT technologies We ve had promising results
How Can You Participate In IoTivity? IoT application developers Open-source contributors Propose new framework features, use cases https://www.iotivity.org/get-involved IoTivity Mailing List
Resources IoTivity SDK and Samples https://www.iotivity.org/ Open Interconnect Consortium http://openinterconnect.org meta-oic Yocto Layer https://git.yoctoproject.org/cgit/cgit.cgi/meta-oic/about/ Working with kernels in the Yocto Project: Presentation https://www.yoctoproject.org/sites/default/files/devdaykernel-tzanussi-elc-2013.pdf Yocto Eclipse IDE Plug-in: Instructional video http://www.youtube.com/watch?v=3zlou-glsh0
Thanks for your time! Q&A